Google’s made a pretty big splash with its AI-generated podcast feature , and before the year is out the app is . As part of a larger redesign of Google’s AI notebook tool, Audio Overviews are now interactive.
After generating an Audio Overview based on the sources you’ve uploaded, Google says you’ll be able to play the recording in a new “Interactive mode (BETA).” Clicking “Join” at any point in that new playback screen will get the AI hosts to call on you to ask a question, which they’ll answer live while you’re listening back. Google cautions that the feature is still experimental and that hosts might pause awkwardly or introduce new inaccuracies while answering questions, but it seemed to work well in a brief test. I was able to create a NotebookLM project trained on articles about NotebookLM, and while asking a question did seem to slow the whole Overview down, the AI hosts were able to smoothly incorporate an answer into the rest of the show.
Alongside these new expanded features, NotebookLM is getting a bit of a visual overhaul. The interface is now split into three sections, a “Studio” panel where AI-generated content like Audio Overviews, study guides, and FAQs live, a central “Chat” panel for asking questions about your sources to Google’s AI, and a “Sources” panel on the left for managing what sources NotebookLM pulls from. It’s a pretty clean setup, and being able to collapse a panel when you’re not using it keeps things from getting cluttered.
Google is also using these updates as a way to introduce its first pass at monetizing NotebookLM. A new premium subscription is available to Google Workspace and Cloud customers as a Gemini add-on, and will give you the ability to generate up to 20 Audio Overviews per day, create up to 500 AI notebooks, and add up to 300 sources per notebook. That translates to an additional $20 per user per month for Workspace subscribers. Starting next year, NotebookLM Plus benefits will also be rolled into the Google One AI Premium subscription.
NotebookLM started as an internal Google experiment called Project Tailwind, but quickly blossomed into one of the more reasonable applications of Google’s Gemini AI model thanks to its grounding in sources you upload, rather than the web and whatever scraped material Gemini was originally trained on. It’s capable of working with anything from web articles to , but its Audio Overviews have proven to be one of its most popular features.
